Anthropic Launches Claude Tag: An Always-On AI Teammate Inside Slack

Anthropic announced "Claude Tag," a new feature that turns its assistant Claude from a single-user chatbot into an always-on "teammate" inside Slack channels. The idea is simple: once Claude is added to a channel and granted specific permissions, any team member can summon it by typing @Claude and assigning it a task, which it completes and returns in the same thread. The feature launched on June 23, 2026, as a research preview for Claude Enterprise and Claude Team customers, and runs on the Opus 4.8 model.

From "Single Assistant" to "Shared Teammate"

What distinguishes Claude Tag from previous integrations is that it works as a shared AI resource for the whole team, not a single user. All channel members interact with one "Claude identity," meaning everyone can see what Claude is working on and continue a task a colleague started from where it stopped. Slack's general manager described this shift as making AI "multiplayer": instead of a private back-and-forth, Claude appears in the open where the team already works.

Persistent Memory and "Ambient" Behavior

Claude Tag adds a layer of persistent memory and context that was hard to maintain in previous tools. The more Claude follows its channel, the more it learns about the nature of the work, and it can gather information from across the organization if granted permission to read other channels. But the most notable feature is "ambient" mode: proactive behavior that makes Claude intervene on its own to keep the team updated, remind of forgotten threads, and follow up on stalled tasks, rather than waiting to be asked. This is precisely what separates it from a traditional chatbot: it monitors channels and steps in when it judges the team would benefit from a reminder, a summary, or a piece of context from another part of the company.

Asynchronous Work

Claude Tag is designed for asynchronous work. Instead of waiting for an instant reply, a user can assign a task and step away, finding it done later. Claude breaks large tasks into stages and executes them with the tools available to it, and can even schedule follow-up tasks and work on projects spanning hours or days. As evidence of the tool's maturity, Anthropic says about 65% of its product team's code is now generated through its internal version of Claude Tag.

Security and Governance: A Pivotal Point

Because a tool with such permissions raises obvious security questions, Anthropic focused on administrative controls. System administrators precisely define which channels, tools, and data sources Claude can access, and its memory and permissions remain "scoped" to the channels designated for it. In practice, this means a Claude instance configured for legal work cannot leak information into the engineering channel. Administrators can also set spending limits at the organization and channel level, and view a full log of Claude's activity, including who initiated each request.

The Competitive Context and Migration

Claude Tag arrives amid a fierce race over "organizational context" as the battleground for enterprise AI. It faces tools like Salesforce's Slackbot, Microsoft's Work IQ, and startups like Viktor and Glean. But Anthropic's approach is narrower and more disciplined: instead of building a horizontal intelligence layer across all applications, it plants AI in the surface where most knowledge work already happens — the team chat. Notably, Claude Tag will replace the current "Claude in Slack" app, with a migration window for administrators and trial credits for eligible customers. Anthropic plans to expand the feature later to platforms beyond Slack.

For Arabic teams that rely on Slack, this opens a practical door to delegating recurring tasks without leaving the work environment, but it also calls for awareness in precisely configuring permissions before granting a proactive tool access to channels carrying sensitive information.